Precertification of low risk digital products by FDA

Occasionally I am asked why I use Twitter. “Why do you need to know what people have had for breakfast? Why get involved with all those crazies?”. I always answer that it’s easy to avoid the bores and trolls (all the easier if one is a straight, white male I suspect) and Twitter is a fantastic way of keeping in touch with interesting people, ideas and developments.

A good recent example was this short series of tweets from Griffin Jones.
This was the first I’d heard of the pre-certification program proposed by the Food and Drug Administration (FDA), the USA’s federal body regulating food, drugs and medical devices.

Griffin is worried that IT certification providers will rush to sell their services. My first reaction was to agree, but on consideration I’m cautiously more optimistic.

Precertification would be for organisations, not individuals. The certification controversy in software testing relates to certifying individuals through ISTQB. FDA precertification is aimed at organisations, which would need “an existing track record in developing, testing, and maintaining software products demonstrating a culture of quality and organizational excellence measured and tracked by Key Performance Indicators (KPIs) or other similar measures.” That quote is from the notification for the pilot program for the precertification scheme, so it doesn’t necessarily mean the same criteria would apply to the final scheme. However, the FDA’s own track record of highly demanding standards (no, not like ISO 29119) that are applied with pragmatism provides grounds for optimism.

Sellers of CMMi and TMMi consultancy might hope this would give them a boost, but I’ve not heard much about these in recent years. It could be a tough sell for consultancies to push these models at the FDA when it is wanting to adopt more lightweight governance with products that are relatively low risk to consumers.

The FDA action plan (PDF, opens in new tab) that announced the precertification program did contain a word that jumped out at me. The FDA will precertify companies “who demonstrate a culture of quality and organizational excellence based on objective criteria”.

“Objective” might provide an angle for ISO 29119 proponents to exploit. A standard can provide an apparently objective basis for reviewing testing. If you don’t understand testing you can check for compliance with the standard. In a sense that is objective. Checkers are not bringing their own subjective opinions to the exercise. Or are they? The check is based on the assumption that the standard is relevant, and that the exercise is useful. In the absence of any evidence of efficacy, and there is no such evidence for ISO 29119, then using ISO 29119 as the benchmark is a subjective choice. It is used because it makes the job easier; it facilitates checking for compliance, it has nothing to do with good testing.

“Objective” should mean something different, and more constructive, to the FDA. They expect evidence of testing to be sufficient in quality and quantity so that third parties would have to come to the same conclusion if they review it, without interpretation by the testers. Check out Griffin Jones’ talk about evidence on YouTube.


Incidentally, the FDA’s requirements are strikingly similar to the professional standards of the Institute of Internal Auditors (IIA). In order to form an audit opinion auditors must gather sufficient information that is “factual, adequate, and convincing so that a prudent, informed person would reach the same conclusions as the auditor.” The IIA also has an interesting warning in its Global Technology Audit Guide, “Management of IT Auditing“. It warns IT auditors of the pitfalls of auditing against standards or benchmarks that might be dated or useless just because they want something to “audit against”.

So will ISO, or some large consultancies, try to influence the FDA to endorse ISO 29119 on the grounds that it would provide an objective benchmark against which to assess testing? That wouldn’t surprise me at all. What would surprise me is if the FDA bought into it. I like to think they are too smart for that. I am concerned that some day external political pressure might force adoption of ISO 29119. There was a hint of that in the fallout from the problems with the US’s Healthcare.gov website. Politicians who are keen to see action, any action, in a field they don’t understand always worry me. That’s another subject, however, and I hope it stays that way.

Advertisements

Quantum computing; a whole new field of bewilderment

At primary school in London I was taught about different number bases, concentrating on binary arithmetic. This was part of an attempt in the 1960s to bring mathematics up to date. The introduction of binary to the curriculum was prompted by the realisation that computers would become hugely more significant. It was fascinating to realise that there was nothing inevitable or sacrosanct about using base 10 and I enjoyed it all.

When I started working in IT I quickly picked up the technical side. I understood binary and with a bit of work, keeping a clear head, everything made sense. It was mostly based on Boolean logic. Everything was true or false, on or off, pass or fail, 1 or 0.

Now this simple state of affairs applied only to the technology. When you started to deal with humans, with organisations, with messy social reality it was important to step back from a simple binary worldview. You can’t develop worthwhile applications, or test them, if you assume that the job simply requires stepping through a process that is akin to a computer program, with every decision being a straightforward binary, yes/no, pass/fail. I’ve talked about that at length elsewhere, eg here “binary opinions – yes or no?“.

Recently I’ve been reading about quantum computing. I’d vaguely known about the subject before. I knew it was a radically different approach to computing, but I hadn’t thought through just how radical the differences are, or the implications for testing. It’s not just testing of course. When a completely new form of computing comes on the scene, one that leaves all our expectations, assumptions and beliefs in tatters, there will be huge ramifications throughout IT. Traditional computers won’t vanish; there will still be plenty of jobs working with them. People will continue to specialise in specific areas of computing. The problem for testers will be that quantum computers are going to be used for new applications, to do things that were previously impossible, and traditional testing techniques won’t necessarily be readily transferable.

What’s so different about quantum computing?

I’m not going to try and provide an introduction to quantum computers. If you think you understand them and you haven’t done some serious study of quantum physics then you’re kidding yourself. And as Richard Feynman said;

“If you think you understand quantum physics, you don’t understand quantum physics.”

If a primary school introduction to binary arithmetic allowed me to get to grips with traditional, digital computers, then the equivalent for quantum computers would be at least an undergraduate degree in physics. Here is a good overview.

I’m just going to run through three features of quantum computers that make my head spin, that have persuaded me that everything I knew about computers will be useless and I will be as well prepared as a peasant trying to get to grips with Excel after stumbling through a time portal from the Middle Ages.

First, instead of bits we have qubits. Bits can be 0 or 1. Once set to a value they don’t change until we perform some operation. The possible states of a qubit are excited or relaxed, which could be considered equivalent to 0 or 1. The fundamental difference is that they can be both at the same time, or rather a mixture of the two. This property is superposition. However, when the qubit is measured it will be frozen as one or the other.

The second weird feature of qubits is that different qubits influence each other. In traditional computing if we operate on a bit then no other bits will change. If they do it’s because the logic of the program controls these changes. In quantum computing that doesn’t apply. Changing a qubit will cause other qubits to change. Qubits are not independent of each other; they are all linked to each other. This is called entanglement.

The third weird thing about quantum computing is that algorithms are probabilistic, not deterministic. A classical digital computer will run through an algorithm and produce the right answer, or rather it will produce a predictable answer that is consistent with the algorithm’s logic. A quantum computer won’t always give the right answer straight away. It might have to send the same input through the same process many times and the results should converge on the right answer, probably.

I’m telling you this not because I think it will give you any sort of useful understanding of quantum computing. A spot of humility is in order. I pass on this information in the hope you realise you have as little idea as I do.

In summary, quantum computing is moving way beyond dear old familiar Boolean logic. Boolean algebra is still valid, but it is only relevant to one part of a wider reality and quantum computers will allow us to explore beyond the Boolean boundaries.

I have done my share of white box testing, delving into the guts of applications to understand what is going on. I know I will never be in a position to do that with quantum computers, even assuming that they are in widespread use before I retire. I doubt if there are many, if any testers, who will be better placed than I am.

What will quantum computers be used for?

If white box testing poses a tough challenge what about black box testing? As I said earlier, quantum computers will be used for problems that traditional computers can’t help with. One application will be the transformation of cryptography. Quantum computing will probably render existing cryptographing techniques obsolete. Other likely uses will be new ways to search and interrogate massive databases, and financial modelling that requires fiendishly intricate calculations on huge volumes of data. These are the applications that caught my eye, because of my background in financial services. A big problem for testers will be, how do you evaluate the output if the computer is doing something that can’t be replicated by other means? This isn’t a new problem, but quantum computers will massively increase the difficulty.

Blind computing guiding the blind?

I wrote about the problems I’d encountered working with big data a few years ago. Our approach could be summed up simply as;

  • build a deep understanding of the domain and the data, and the essential relationships or rules,
  • get the business or domain experts heavily involved and pick their brains,
  • ensure you influence the design so that it is testable, ideally so that the solution enables testing of separate segments which can be isolated so you can bring your acquired knowledge and understanding to bear.

The first two points will always be relevant regardless of the technology. The third one has been picked up by quantum computing scientists who have developed a technique called blind quantum computing.

The blind technique involves running a separate, simpler(!), quantum computer called a verifier alongside the server running the test application. The verifier feeds the server with small, discreet calculations to perform. It does this in a way so that only the verifier, and the testers, can know what the answer should be, and the server knows only what operations it should perform. This is the simplest explanation of blind quantum computing I’ve been able to find.

An obvious observation is that this is hardly black box testing as we have known it. Who programs and controls the verifier? It is a quantum computer, and I doubt if ISTQB accreditation will ever get you very far here.

Fields of bewilderment

I don’t have any answers here, and I strongly suspect easy answers will never be available to testers who work with quantum computers. That, perhaps, is the point to take away from my article. Writing this has felt less like trying to spread some understanding of an immensely difficult subject, and more like an account of a bewildered ramble through my freshly expanded fields of ignorance.

I can’t state with confidence that software testers will ever move into quantum computing. Perhaps the serious testing will be done by quantum computing scientists. That raises its own dangers. Will they have the right instincts, an appropriate sceptical and questioning approach? There is already talk of quantum computers permitting the development of perfect software. Apparently they will be able to work their way through every conceivable permutation of data to establish that the application was built to its specification. Of course there is a huge, familiar old question being begged there; how does anyone know the spec was right?
Opening of the poem 'Youth and Hope' from which the quote 'perplexity is the beginning of knowledge' comes

I worry that good testers may be frozen out of working with quantum computers, and that the skills and questions they do have, and that will remain relevant, will be lost. Whatever happens I am confident that testers imbued with the principles of Context-Driven Testing will be better prepared than those who know only what they learned to pass ISTQB exams, or who are accustomed to following prescriptive standards. Whatever the future holds for us it will be interesting! After all, perplexity is the beginning of knowledge.

Presentations

This page will carry slide decks for presentations I’ve given at conferences – as I get round to adding them. All the presentations are in PDF format and open in new tabs.

Farewell to “pass or fail”?

I gave this presentation, “Farewell to ‘pass or fail’” at Expo:QA 2014 in Madrid in May 2014. It shows how auditing and software testing have faced similar challenges and how each profession can learn from the other.

Standards – promoting quality or restricting competition?

This presentation is the one that attracted so much attention and controversy at CAST 2014 in New York. It is a critique of software testing standards from the perspective of economics.

ISO – the dog that hasn’t barked

On August 12th I gave a talk at CAST 2014, the conference of the Association for Software Testing (AST) in New York, “Standards; promoting quality or restricting competition.” It was mainly about the new ISO 29119 software testing standard, though I also wove in arguments about ISTQB certification.


I was staggered at the response. Iain McCowatt asked what we should do in response. Karen Johnson proposed a petitition, which subsequently became two. Iain set up a petition through the International Society for Software Testing (ISST), directly targetted at ISO.

Karen’s petition is a more general manifesto for all professional testers to sign if they agree with its stance on certification and standards.

I strongly commend both the petition and the manifesto to all testers.

Eric Proegler also set up an AST special interest group to monitor and review the issue.

This action was not confined to the conference. In the last three weeks there has been a blizzard of activity and discussion on social media and throughout the testing community. Many people have blogged and spoken out. I gave a brief interview to Infoworld.com, and wrote an article for the uTest blog.

Huib Schoots has drawn together many of the relevant articles on his blog. It’s a valuable resource.

However, my own blog has been embarrassingly silent. I’ve not had time, till now, to write anything here. I’ve seen a big rise in traffic as people have hunted down articles I have previously written. August was the busiest month since I started the blog in 2010.

Significant and sustained opposition

Finally I have had a chance to set some thoughts down. I never expected my talk to get such a huge reaction. The response has not been because of what I said. It has happened because it came at the right time. I caught the mood. Many influential people, whom I respect, realised that it was time to speak out, and to do it in unison.

There has been significant and sustained opposition to ISO 29119 as it has developed over the years. However, it has been piecemeal. Individuals have argued passionately, authoritatively and persuasively against the standard, and they have been ignored.

The most interesting response since my talk, however, has been from the dog that didn’t bark, ISO. Neither ISO nor the 29119 working group has come out with a defence of the standard or the way that it was developed.

They have been accused of failing to provide a credible justification for the standard, of ignoring and excluding opponents, and of engaging in rent-seeking by promoting a factional interest as a generic standard.

And their response? Silence. In three weeks we have heard nothing. In their silence ISO have effectively conceded all the points that the Stop 29119 campaigners have been arguing.

There have been some sporadic and entirely unconvincing attempted defences of the standard, mixed up the with odd spot of risibly offensive ad hominem attacks. Collectively the weakness of these defences exposes ISO 29119 for the sham that it is.

Defence #1 – “It’s a standard”

This is possibly the weakest and most wrong-headed argument. ISO are trading on their brand, their image as a disinterested promoter of quality. It is disappointing how many people accept anything that ISO does at face value.

It is important that promoters of any standard justify it, demonstrate that it is relevant to those who will have to use it and that they enjoy a consensus amongst that community. In practice, ISO can spin the argument around.

Once ISO anoints a document with the magic word “standard” too many people suspend their own judgement. They look for opponents to justify their opposition, with reference to the detail they object to in the standard.

In the absence of such detailed arguments against the standard’s content people feel content with saying “it’s a standard, therefore it is a good thing”. That argument might impress those who know nothing about testing or ISO 29119. It lacks any credibility amongst those who know what they are talking about.

Defence #2 – “It’s better than nothing”

Whoops. Sorry. I have to backtrack rapidly. This argument is even worse than the last one. Something that is lousy is emphatically not better than nothing. Just doing something because it’s better than doing nothing? Oh dear. It’s like being lost in a strange city, without a map, and putting a blindfold over your eyes. Well, why not? You’re lost. You don’t know what you’re doing. Blindfolding yourself has to be better than nothing. Right? Oh dear.

This is the politicians’ fallacy, as explained in the classic comedy “Yes Minister”. Thanks to Scott Nickell for reminding me about it (see comments below).


If an organisation’s testing is so disorganised and immature that ISO 29119 might look like the answer then it is ripe ground for a far better approach. Incompetence is not a justification for wheeling in ISO 29119. It is a justification for doing a better job. Next argument please.

Defence #3 – “ISO 29119 doesn’t have to be damaging. It’s possible to work around it.”

The arguments in favour of ISO 29119 aren’t really getting a whole lot more convincing. Sure, you might be able to mitigate some of the worst effects by tailoring your compliance, or ignoring parts that seem particularly unhelpful. However, that is putting effort into ensuring you aren’t worse off than you are if you do nothing.

Also, if standards and rules are so detailed and impractical that they cannot be followed in full then it exposes people to arbitrary enforcement. Once users start tailoring the standard they will leave themselves open to the accusation that they did not comply in full. If things go well there will be no thanks for tailored compliance.

If there are problems, and there are always problems, then any post mortem will reveal that the standard wasn’t followed in full. The implication will be that the problems were caused by deviation, not that the deviation was the only reason anything worthwhile was achieved. Testers and developers will rightly fear that retribution will be arbitrary and quite unrelated to the level of care and professionalism they brought to their work.

Further, ISO 29119 is being marketed with an appeal to fear, as a way of protecting individuals’ backsides if things go wrong. If managers buy into the standard on that basis, are they really likely to take a chance with tailoring the standard?

ISO 29119 is also being pushed to buyers who don’t understand what good practice in testing means. Are they not likely to play safe by insisting in contracts and internal standards that ISO 29119 is complied with?

No, the possibility that we can “work around it” is not a credible argument in favour of ISO 29119.

Defence #4 – “The dissenters are just self-interested”

The standard is apparently a threat to our “craft” and to our businesses. Well, there’s more to it than that, but even if that were the only objection it would still be an entirely valid one. We have argued strenuously that the standard is anti-competitive. A riposte that we fear it will hit us financially is merely to concede the argument in a graceless way.

Anyway, if I were chasing the money I could have made a lot more, an awful lot more by taking lucrative test management contracts to do poor quality work managing the testing process on dysfunctional projects.

I can do great documentation. I am literate, organised and have the knack of getting tuned into corporate politics. My ability to churn out high quality, professional looking documents that would get the money in from clients was a great asset to my last employer.

Testing? Sorry, I forgot about that. I thought we were talking about documentation and money.

Defence #5 – “They’re whingers who wouldn’t get involved.”

This point ignores the argument that the ISO process for developing standards is carefully designed to exclude those who disagree. Again, the “whingers” line spins the argument around. It is not our responsibility to justify our failure to get involved with rent seekers and try to persuade their committees to desist.

I have seen the ISO 29119 working group’s schedule for meetings worldwide over the last few years. In the unlikely event that I would have been allowed to join, my expenses would have wiped out a huge slice of my income. It would certainly have cost me far more than we’ve spent on family holidays in that period. And for what? I’d have sacrificed all that time and money in order to be ignored. Those gallant souls who have tried to fight from the inside have been ground down and spat out by the system. That’s how the system works, how it is intended to work.

No, I didn’t get involved. I had better things to do with my time. In any case, it seems that my campaigning from the outside has not been ignored!

So what now?

We keep going. All this activity has just been the start. ISO is not going to withdraw the standard. However, the blogs, articles and petititions lay down a marker. It shows that the standard was not developed according to any plausible definition of consensus, and that it lacks credibility.

The opposition will strengthen the resolve of testers who wish to stop their organisations buying into the standard. It will provide ammunition to those who want to persuade lawyers and purchasing managers from requiring compliance to ISO 29119.

This one is going to run. We are not going away, and if ISO continue to ignore us then they will simply make themselves look foolish. Do they care? Or are they confident they have enough power and can just sail on regardless?

The problem is not the problem

There are two issues that have been bugging me for a while now and I mean to get them off my chest this week.

They’re not software testing topics but as with so many psychological, organisational or social problems they do have a very real relevance for testers. We don’t exist in a technical or procedural bubble that insulates us from the blizzard of conflicting and confusing influences that swirl around every organisation.

The first issue is that of problems. Very often the problem is not the problem. The problem we see is not necessarily the real, possibly very damaging, problem.

As testers this shouldn’t be news. We’re used to root cause analysis, delving behind the curtain to find out why things have gone wrong and not being deceived by what we initially see. We should also be familiar with the “five whys”, or some variant on that; the discipline of repeatedly challenging answers till we burrow down to the heart of the problem.

That’s not what I want to talk about here. Things go wrong, in life, at work, and especially in software development. That’s the way of it.

What really matters, certainly in the long run, is often not the problem itself but the way we deal with it.

We can deal with the problem in a brisk and efficient manner, resolving the immediate concerns, learning the right lessons and moving on. Or we can blunder into any one of several traps that could hurt us far more than the original problem did.

The first is the cover up. Watergate is the great historical example. The original crime might have been handled without damaging Richard Nixon, but it was the cover up that ended his presidency.

When we, as software developers, hit problems the worst thing we can do is attempt to cover up. Honesty with customers and users isn’t simply honourable and ethical, it also buys you trust and support.

Own up on the smaller problems and it may just buy you the credibility and goodwill to save your skin when you hit big, career threatening problems.

Denial is the sibling of the cover up. It might not be a vigorous, outspoken denial. It may just be a matter of kidding ourselves that there’s nothing really wrong, that we can carry on as before once we’ve sorted the problem. It was just bad luck, or a bit of carelessness. No big deal. Just move on folks. Nothing to learn here.

The next trap is the one you plunge into if you head off in the opposite direction; over-reaction. Losing your temper and savaging staff is obviously damaging and isn’t tolerated in healthy organisations any more, but a more insidious variant is still too prevalent. That is the tendency to blame individuals.

Someone must be to blame, somebody must have screwed up and be held to account. I once worked somewhere with an appalling blame culture. Problems were something to run away from, not something to fix. No-one wanted to jump in and proactively resolve problems in case they got implicated in the management investigation and fallout.

A more respectable form of over-reaction is the assumption that the problem justifies sweeping changes or drastic action rather than careful tweaking. Instead of a careful analysis of what actually went wrong and what the cause was, there is the presumption that it was confirmation of a trend, or deeper problem, that was previously suspected.

That leads me on to the fourth trap, confirmation bias. We convince ourselves the world should work a certain way. Any and every problem can be interpreted as supporting evidence.

The glaring example in the history of software development was the conviction that software development should be essentially the same as construction engineering. Early experts saw practitioners struggling to manage projects with a more or less structured and orderly linear process. They concluded that the answer to the problem was to try and make the process even more rigidly structured, more ordered and more linear. We got Structured Methods (shudder!).

After all, nobody builds a bridge incrementally, opening it bit by bit. No-one erects a building in several iterations. So software engineering shouldn’t be any different. That was more or less the line of thought. Life might be simpler if software development were a highly predictable and controllable process, but the truth is that development is not like that.

I referred to this problem on my blog earlier this year when I explained the predictability paradox.

The correct response to the problem would have been to go with the grain of the problem, embracing the uncertainty, building iteration and incremental development into the development process rather than pretending that we understand the business problem sufficiently well to proceed down a rigid, linear route.

The problem was assumed to be that software engineering was not sufficiently like construction engineering. Divergence between the two activities was seen not as evidence of a genuine difference, but as proof that the software developers were out of line. The truth was that software development is a quite separate discipline and the real problem was that developers were forced to contort their practices to fit a wildly inappropriate paradigm.

The real problem wasn’t the problem the profession saw. It was their response to the problem.

None of this is new to marketing people. Apparently BMW owners who have a problem with their car are more likely to buy another BMW than those who have no problems. Why? It’s because of the way that BMW deal with problems. Is it true? Well, I don’t have access to the statistics, but that’s BMW’s reputation. Problems have become an asset, rather than remaining a problem.

What does this mean for any of us? It’s not easy to draw a neat solution to this problem of problems. All I can pass on is something I have found useful.

Often we plunge into one of the traps because we choose to respond in a way that suits our instincts or prejudices, or which is just the easy, comfortable response.

Take a breath before committing yourself and ask yourself whether your response is what ought to happen. Or is it the response that you’d really like to be the right one? Is it the easy one? Are you maybe dodging the real issue, going for short term expediency rather than long term results?

Run through the traps I’ve discussed. Do any of these make you wonder about your response?

Is your reaction to the problem going to be seen as the real problem further down the line?

After I wrote this I checked the phrase “the problem is not the problem” to see if anyone had used it before. Initially I was a bit disappointed to see that Tom Peters has, but then I cheered up when I realised he’s a pretty respectable ally! Watch the video in the link or, if you’re in the office, check out the transcript (PDF, opens in a new window). It’s very much in line with my argument, but he puts a bit of extra spin on it.

The other thing I want to get off my chest this week is about making decisions. I’ll be back in a day or so.

Please add any comments. I do like to see what people think.

The picture in the header

This was taken in Glen Shirra in Inverness-shire, looking north-east over Loch Crunachdan and upper Strathspey; Ordnance Survey map reference 540918.

I took this photo on my cycle ride from Perth to Inverness in June 2009. That day was one of the best of my trip. I’d taken the train from Rannoch Station to Corrour Halt, then cycled through Strathossian and along Loch Laggan. I took the short and scenic, but very rough, route over the top from Aberarder Lodge and down into Strathspey.

I was on a high when I took this picture. After slogging up the steep track from Loch Laggan I was now bumping downhill into Strathspey in glorious warm sunshine through fantastic scenery. Scotland was looking at its very best that day and it was great to be alive.